MSDP Workshop PRACTICAL SESSION P. Maeght, P. Mein N. Mein, M. Lafon, N. Meunier T. Roudier MSDP Workshop, Tarbes, 18-20th January 2006
Directories oceanide /data3/DPSMWS/day/ : raw sequences /data4/groupX/ (one dir. / group, X=1,2) auto0503/ : code for THEMIS observations until 2004 auto0601/ : code for THEMIS observations since 2005 dalsa/ : code for VTT docdpsm/ dpsm_form/ Results will be in auto0503/ or auto0601/
Directories gaia /baie1/DPSMWS/day/ : raw sequences /baie1/groupX/ (one dir. / group, X=3,4) auto0503/ : code for THEMIS observations until 2004 auto0601/ : code for THEMIS observations since 2005 dalsa/ : code for VTT docdpsm/ dpsm_form/ Results will be in auto0503/ or auto0601/
+ 0503
VTT processing Similar algorithm but manual (no msdpauto, no sequence.par) editing of dalsa.pro, ms.par (file time …) Start ms1 directly
In autoXXXX/ The msdpauto (IDL) will : create directories for the results dir_date_seqnum_camnum/ copy the code in it go to this directory build the ms.par start the fortran code put the results there Copy the resulting g* file (geometry) in the upper directory in case useful for another sequence
Goto /data4/groupX/ or /baie1/groupX/ Goto the autoxxxx/ corresponding to the sequence Make the sequence.par if not provided manually (see example in /data3/DPSMWS/sequence_model.par) automatic with the BASS2000 interface To get bs + camera number : edit fits file x=mrdfits(/data…./*.fts’,ran=[0,0]) (WARNING 2004 : 2 cameras / line) Depending on the case : start msdpauto, « /dir_data/ » start msdpauto, « /dir_data/ », /no_fort, modify ms.par then start ms1 other procedure if indicated File vizualization How to process a sequence
File vizualisation IDL : x=readmsdp(‘file’) (q* or p* file) check the image order in the file tvscl,x(*,*,i) ; plot… (cut) etc. When using the BASS2000 interface : choose q or p, and « Voir DEP » : equivalent to tvscl Postscript files : kghostview file.ps ; or « Voir PS » in the BASS2000 interface
Rep auto Make sequence.par msdpauto Edit sequence.par if modification If modif ms.par
Categories of sequences Good sequences (THEMIS, VTT) Sequences with interruption during the processing Sequences providing maps or profiles showing problems (visual verification)
For each sequence Date, sequence number Line(s) Volume Time to process (for 100% CPU time on our machine) Problem if any, illustration before / after
To reduce processing time after transformation b3 to individual files : tob1, tob2 in ms.par ; or nob1, nob2 (msdpauto should have been started with /no_fort) stop at « q » step (qp = 0 in sequence.par) for quicklook only (no profile) keep only one stokes if several are observed (qv, nq in sequence.par) ; see SEQ_STOKES in fits header Beam exchange : nqseul = 7, 8, 9 in ms.par (otherwise 0)
Sequence : #1, 9 aug 2004 Line: 5896 Volume : 98 Mo Seq_stokes : +V Processing time : <16 min No problem
Sequence : #1, 20 oct 2003 Line: 5896 & 8542 Volume : 2.2 Go Seq_stokes : +Q+U+V Processing time : <28 min / line No problem Use nq=3, qv=3
Sequence : #3, 20 oct 2003 Line: 8542 (cam b01) Volume : 5.9 Go (2 lines) Seq_stokes : +Q+U+V Processing time : <1h30 No problem Use nq=3, qv=3
Sequence : #7, 31 oct 2003 Line: 5896 & 8542 Volume : 1.7 Go Seq_stokes : +V+Q+U Processing time : <35 min / line No problem Use qp=3, qv=1
Sequence : #6, 25 oct 2002 Line: 5896 (cam b02) Volume : 3.9 Go Seq_stokes : +V Processing time : <1h20 No problem
Sequence : #3, 25 oct 2002 Line: 5896 Volume : 1.6 Go Seq_stokes : +V Processing time : <1h30 No problem ; illustration dt=0,60 Use tob1, tob2 dt=60
dt=0
Sequence : #7, 6 oct 2005 Line : 5896 Volume : 660 Mo Seq_stokes : +V Processing time : 20 min dt/ntmax problem + intert dt=60 in sequence.par
dt=0 in sequence.par ; or ntmax=6, intert=0 in ms.par
Sequence : 17 sept 2005, VTT Line: Halpha, no polar Volume : 2.7 Go No problem
Sequence : #2, 31 jul 2004 Line : 5896 Volume : 300 Mo Seq_stokes : +V Processing time : 16 min milgeo problem, must use seq #5 g* file It is also possible to impose #seq calib in sequence.par Seq 5, 9:26 Seq 2, 7:57
Sequence : #12, 2 aug 2004 Processing time : 20 min milgeo problem, illustrate very large value of milgeo
Sequence : #14, 31 jul 2004 Processing time : 15 min illustration bad data
Sequence : #13, 13 oct 2003 Processing time : 18 min illustration bad data
Sequence : #4, 14 oct 2003 Processing time : 1h05 illustration bad data
Sequence : #3, 15 oct 2003 Processing time : 48 min illustration bad data
Sequence : #11, 10 aug 2004 Processing time : 23 min Keyword problem : polarization or not ?
Sequence : #3, 15 oct 2004 Line : 5896 Volume : 287 Mo Seq_stokes : +V-V Processing time : 1h03 min for whole seq inveri problem +xfirst + illustration tob1/tob2 Use ilisqp=3, jlisqp=3 or more to smooth inveri=0
inveri=1
inveri=1, tob1= , tob2=
Sequence : #3, 16 oct 2002 Line : 5896 Volume : 1.3 Go Seq_stokes : +V Processing time : 2h10 min whole scan xfirst problem + norma + illustration tob1/tob2 xfirst=0
xfirst=1
xfirst=1,norma=0199
Sequence : #11, 28 oct 2002 Line : 5896 Volume : 1.1 Go Seq_stokes : +V Processing time : ? Small maps with scan problems dt=60, xfirst=0
dt=0, xfirst=0
dt=0, xfirst=1